CN1567331A - Method for generating electrocortical potential nonlinear trend diagram used in real-time monitoring - Google Patents

Abstract

This invention discloses a kind of generating method of real-time monitored brain potential trend chart by adopting computer technique and non-linear dynamics analysis method. It relates to a kind of generating method of brain potential non-linear trend chart. It comprises the following steps: first, collecting original EEG information from different position of scalp; A/D converting to digital EEG information; according to the non-linear dynamics theory and by adopting the trail of state space in time sequence reconstruction system, within one second, converting the conductive digital EEG information to non-linear index that relating to dimension, complexity and approximate entropy, which reflected the current non-linear dynamics characteristic; drawing the non-linear trend chart by using the non-linear index of time sequence, to reflect the non-linear dynamics characteristic variation condition of entire course. This invention can real-time monitor the anaesthesia effect, reflect the variation of anaesthesia effect accurately and promptly.

Description

The generation method that is used for real-time monitored brain potential nonlinear trend figure

Technical field

The invention discloses a kind of generation method of utilizing computer technology and nonlinear dynamic analysis method to realize being used for real-time monitored brain potential nonlinear trend figure.It relates to a kind of brain potential nonlinear trend map generalization method.

Background technology

Tradition eeg analysis means are concluded and can be divided into two big classes: time-domain analysis and frequency-domain analysis.Time-domain analysis comprises wave amplitude, frequency, time-histories, Transient distribution and area under curve etc., mainly reflects the geometric properties of EEG signals.Frequency-domain analysis performance EEG signals data are with the rule of frequency change and the information that is reflected, its core be each frequency band power spectrum estimation based on (fast) Fourier transform, comprise power spectrum, be concerned with etc.Fourier transform require signal be determine and stably, and exist signal many bursts, transient state (as spike etc.) in the EEG signals, the analysis of spectrum reflection is relatively poor in this case.In recent years, utilize nonlinear kinetics principle such as chaos and fractal theory and method to come the functional activity state of research and analysis brain, become the new focus of cerebral function research.Studies show that: the EEG signal originates from the nonlinear system of a height; Electrical activity of brain has the determinacy chaotic characteristic.That nonlinear dynamic analysis can provide is that linear analysis can not obtain, the information of relevant neural network function, and the situation of brain function activity variation track can be provided.Therefore the Nonlinear Dynamics cognitive change information of consciousness that is more suitable for monitoring in the electrical activity of brain and comprised, it has represented the future thrust of electroencephalogramsignal signal analyzing method.At present,, be difficult to realize real-time analysis, also lack the analytical approach of the long-time process of reflection in real time because nonlinear dynamic analysis need carry out the complex analyses of mass data.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, a kind of generation method of utilizing computer technology and nonlinear dynamic analysis method to realize being used for real-time monitored brain potential nonlinear trend figure is provided.It gathers original brain electric information from the scalp different parts, be converted to the digitizing brain electric information through A/D, according to the nonlinear kinetics principle, the track of operate time sequence reconfiguration system in state space, the digitizing brain electric information that respectively leads is converted to correlation dimension, complexity and the approximate entropy nonlinear exponent of the current nonlinear dynamic characteristic of reflection, this conversion was finished within 1 second, utilize the seasonal effect in time series nonlinear exponent to draw out nonlinear trend figure, reflect the nonlinear dynamic characteristic situation of change of whole process.The present invention can apply to the real-time monitoring of depth of anesthesia in the anesthesia process, for the anesthesia process provides accurately real-time depth of anesthesia situation of change, provides objective basis for the anesthesia doctor takes appropriate measures; Carry out the anaesthetic effect assessment, and the distribution of research anaesthetic in brain, to the big higher brain function situation etc. that exerts an influence.Also can be used for the EEG signal analysis under the situations such as big brain cognitive function activity, different physiological status, various dull-witted state, mental illness and neural rehabilitation.

In order to reach above-mentioned technique effect, the present invention adopts following step: (1) gathers original brain electric information from the scalp different parts; (2) the original EEG signals that will collect is carried out A/D and is converted to digital information, and the various interference of filtering also are stored in the storer; (3) the operate time sequence reconstruct track of EEG signals in state space that respectively lead; (4) in correlation dimension, complexity and three kinds of non-linear eeg dynamic analysis methods of approximate entropy, select a kind of the analysis; (5) brain electric information that is buffered in the storer is carried out the recursion processing, reconstruct comprises the new vector set of recursion information; (6) vector set after handling through recursion is pressed selected non-linear eeg dynamic analysis method respectively, use the method for simplifying and optimizing, carry out the calculating of correlation dimension; Select suitable and optimum parameters, carry out the calculating of complexity and approximate entropy, realize real-time analysis; (7) utilize time series correlation dimension number, complexity and approximate entropy nonlinear exponent, reflect current nonlinear dynamic characteristic, draw out brain potential nonlinear trend figure---nonlinear trend figure in chronological order, reflect the nonlinear dynamic characteristic situation of change of whole process; (8) store and export resulting nonlinear exponent and nonlinear trend figure.

Describedly brain electric information carried out recursion handle the following method of taking:

Before the track in the reconfiguration system state space, carry out recursion and handle being buffered in EEG signals V (i) in the storer, with EEG signals sequence X (i)=(x (1), x (2), x (3) ..., x (N)) handle through recursion after, convert new vector set to:

$V\left(i\right)=\left[\begin{array}{c}X(i-k)\\ X(i-k+1)\\ \·\·\·\\ X\left(i\right)\end{array}\right]$ K=m/N wherein, 0≤k≤i

Wherein m is a recursion length, is through the EEG signals sequence after the recursion processing in each data of handling, and has so just guaranteed the continuity of result; After disposing, the V (i-1) that the original EEG signals sequence of the V in the buffer memory (i) is moved in the buffer memory locates.

Following method is taked in the calculating of described correlation dimension:

Correlation dimension D ₂Calculating:

By the track of time series reconstructing system in state space that is studied: realize reconstruct M dimensional vector V (i) by the method that embeds time delay, L time delay, M is for embedding dimension:

V(i)＝(x(i)，x(i+L)，x(i+2×L)，...，x(i+(M-1)×L))

The wherein calculating of correlation integral:

$C\left(r\right)=\frac{1}{N_{\mathrm{ref}}}\underset{i=1}{\overset{N_{\mathrm{ref}}}{\mathrm{\Σ}}}\frac{1}{N-i}\underset{j>i}{\overset{N}{\mathrm{\Σ}}}\mathrm{\θ}(r-||V\left(j\right)-V\left(i\right)|\left|\right)$

Wherein θ is the Heavside function: if x＜0, θ (x)=0; If x 〉=0, θ (x)=1; N _RefNumber for reference point; Find out the linear segment between logC (r)/log (r), obtain the slope of this partial fitting straight line, this slope is correlation dimension D ₂:

$D_2=\underset{r\→0}{\lim }\frac{\log C\left(r\right)}{\log \left(r\right)}$

Following method is taked in the calculating of described complexity and approximate entropy:

The calculating of complexity is calculated according to the algorithm of Kasper and Schuster;

The calculating of approximate entropy is calculated according to the algorithm of Pincus;

In order to realize real-time processing, the computational data length of complexity and approximate entropy is taked the 250-1000 data point, and selects frontal temporal part to lead and carry out real-time analysis.Like this, make 16 to lead length and can within 0.5 second, finish calculatings (computing machines of use P4 1.7G, 128,000,000 internal memories), reached the requirement of real-time analysis fully for the EEG signals of whenever leading 500 data to employed in calculating.

Following method is taked in the drafting of the generation of described nonlinear exponent and nonlinear trend figure:

The correlation dimension that calculates as stated above, complexity and approximate entropy are according to the recursion disposal route, for R as a result _iHave:

Wherein, the span of n is between m/N～50;

Utilize the nonlinear exponent of the correlation dimension, complexity and the approximate entropy that calculate in real time, reflect current nonlinear dynamic characteristic.Go out brain potential nonlinear trend figure by nonlinear exponent and time relation real-time rendering again, reflect the nonlinear dynamic characteristic situation of change of whole process.

The effect that the present invention is compared with prior art useful is: nonlinear exponent and nonlinear trend figure by the present invention obtains, realized the real-time analysis of nonlinear kinetics; The variation of the nonlinear dynamic characteristic of the EEG signals of surveying can be reflected in real time, the variation tendency of the nonlinear dynamic characteristic of whole observation process can be understood easily; The complex information that helps us to understand in the EEG signals to be comprised reflects the electronic attitude change procedure of brain comprehensively; In addition,, can apply to the real-time monitoring of depth of anesthesia in the anesthesia process,, provide objective basis for the anesthesia doctor takes appropriate measures for the anesthesia process provides accurately real-time depth of anesthesia situation of change by nonlinear exponent and nonlinear trend figure; Carry out the anaesthetic effect assessment, and the distribution of research anaesthetic in brain, to the big higher brain function situation that exerts an influence; Brain function activity is especially realized the Changing Pattern of original brain electric informations such as cognition under the research narcosis.Also can be used for the EEG signal analysis under the situations such as big brain cognitive function activity, different physiological status, various dull-witted state, mental illness and neural rehabilitation.

Description of drawings

Fig. 1 is a workflow synoptic diagram of the present invention.

The peace and quiet EEG signals waveform of closing one's eyes before Fig. 2 (a) anesthesia process begins;

Fig. 2 (b) anesthesia process general anesthesia state hypencephalon electric signal waveform;

Fig. 2 (c) anesthesia recovery process EEG signals waveform;

Fig. 3 (a) anesthesia process begins nonlinear exponent and nonlinear trend figure under the preceding quiet closed-eye state;

Nonlinear exponent and nonlinear trend figure under Fig. 3 (b) anesthesia process general anesthesia state;

Embodiment

The present invention is described in further detail below in conjunction with the drawings and specific embodiments.

Fig. 1 has shown workflow of the present invention.The present invention uses and leads the physiological signal acquisition analysis system more, lays FP according to international standard 10～20 systems of leading at patient's scalp of accepting anesthesia ₁(left front volume), FP ₂(right front volume), T ₃(left temporo), T ₄(right temporo), C ₃(top, a left side), C ₄(right top), O ₁(left side pillow), O ₂(right pillow) 8 conductive electrode, hanging down with ears is reference.Record is accepted the patient of anesthesia from beginning to accept to anaesthetize the overall process that operation finishes the anesthesia recovery.Original brain electric information sample frequency is 500 or 1000Hz, 12 of moulds/number conversion precision.The original EEG signals that collects is carried out A/D be converted to digital information, the various interference of filtering also are stored in the storer, and length is 2000 points.

Before carrying out the track of reconstructing system in state space, the EEG signals V (i) that is buffered in the storer is carried out the recursion processing, with EEG signals sequence X (i)=(x (1), x (2), x (3) ..., x (N)) through after the recursion processing, convert new vector set to:

$V\left(i\right)=\left[\begin{array}{c}X(i-k)\\ X(i-k+1)\\ \·\·\·\\ X\left(i\right)\end{array}\right]$ K=m/N wherein, 0≤k≤i

Wherein m is a recursion length, is through the EEG signals sequence after the recursion processing in each data of handling, and has so just guaranteed the continuity of result; After disposing, the V (i-1) that the original EEG signals sequence of the V in the buffer memory (i) is moved in the buffer memory locates.

Correlation dimension D ₂Calculating: by the track of time series reconstructing system in state space that is studied: realize reconstruct M dimensional vector V (i) by the method that embeds time delay, L time delay, M is for embedding dimension.

V(i)＝(x(i)，x(i+L)，x(i+2×L)，...，x(i+(M-1)×L))

The calculating of correlation integral:

$C\left(r\right)=\frac{1}{N_{\mathrm{ref}}}\underset{i=1}{\overset{N_{\mathrm{ref}}}{\mathrm{\Σ}}}\frac{1}{N-i}\underset{j>i}{\overset{N}{\mathrm{\Σ}}}\mathrm{\θ}(r-||V\left(j\right)-V\left(i\right)|\left|\right)$

Wherein θ is the Heavside function: if x＜0, θ (x)=0; If x 〉=0, θ (x)=1; N _RefNumber for reference point; Find out the linear segment between logC (r)/log (r), obtain the slope of this partial fitting straight line, this slope is correlation dimension D ₂:

$D_2=\underset{r\→0}{\lim }\frac{\log C\left(r\right)}{\log \left(r\right)}$

The calculating of complexity and approximate entropy:

The calculating of complexity is calculated according to the algorithm of Kasper and Schuster;

The calculating of approximate entropy is calculated according to the algorithm of Pincus;

In order to realize real-time processing, the computational data length of complexity and approximate entropy is taked the 250-1000 data point, and selects frontal temporal part to lead and carry out real-time analysis.Like this, make 16 to lead length and can within 0.5 second, finish calculatings (computing machines of use P4 1.7G, 128,000,000 internal memories), reached the requirement of real-time analysis fully for the EEG signals of whenever leading 500 data to employed in calculating.

The correlation dimension that calculates as stated above, complexity and approximate entropy are according to the recursion disposal route, for R as a result _iHave:

Wherein, the span of n is between m/N～50; We use the n value is 2 o'clock, and the calculating of nonlinear exponent has reached the requirement of real-time processing.

Utilize the nonlinear exponent of the correlation dimension, complexity and the approximate entropy that calculate in real time, reflect current nonlinear dynamic characteristic.Real-time rendering goes out brain potential nonlinear trend figure-nonlinear trend figure in chronological order again, reflects the nonlinear dynamic characteristic situation of change of whole process.

Fig. 2 (a), 2 (b), 2 (c) have shown respectively from what the patient who accepts anesthesia collected and have led physiological signal more.Wherein, the first eight leads signal for respectively from scalp FP ₁(left front volume), FP ₂(right front volume), T ₃(left temporo), T ₄(right temporo), C ₃(top, a left side), C ₄(right top), O ₁(left side pillow), O ₂The brain electric information at (right pillow) 8 positions, the 9th leads and is myoelectric information.Fig. 2 (a) for the anesthesia process begin before under the quiet closed-eye state, Fig. 2 (b) under the anesthesia process general anesthesia state, Fig. 2 (c) is anesthesia process end recovery process.Can see that there was difference in quiet closed-eye state before EEG signals and anesthesia process began in the anesthesia process, but more approaching with anesthesia process end recovery process.The complex information that we are difficult to find out the EEG signals and are comprised from naked eyes.Myoelectric information shows that as Fig. 2 (a), body exists myoelectrical activity under waking state, and must to keep suitable muscular tension relevant with body for this; Using the anesthesia process of muscle relaxant, because the effect of medicine has been blocked the contact between nerve-muscle, so we almost can't see myoelectrical activity such as Fig. 2 (b) in the anesthesia process; If but the drug effect of muscle relaxant descends or disappearance, then myoelectrical activity can occur.

Fig. 3 (a), 3 (b) have shown nonlinear exponent and the nonlinear trend figure that accepts the patient of anesthesia respectively.Quiet closed-eye state before on behalf of the anesthesia process, Fig. 3 (a) begin, Fig. 3 (b) has represented anesthesia process general anesthesia state.The numerical value on the left side is nonlinear exponent, has reflected the depth of anesthesia situation of current time; The curve on the right is nonlinear trend figure, has reflected the situation of change situation of whole anesthesia process depth of anesthesia.Fig. 3 (a), 3 (b) have shown that respectively two positions (are specially FP ₁And FP ₂) approximate entropy nonlinear exponent and nonlinear trend figure.Can find current time FP from Fig. 3 (a) ₁And FP ₂The approximate entropy nonlinear exponent be respectively 1.11 and 1.06; The whole peace and quiet process nonlinear trend figure that closes one's eyes shows that nonlinear exponent numerical value fluctuates between 0.8～1.1.This result shows, the approximate entropy nonlinear exponent keeps higher relatively level under the waking state, and this mainly is because cerebral cortex still keeps still having due to a certain amount of communication between to a certain degree active, the cerebral cortex neurocyte.Can find current time FP from Fig. 3 (b) ₁And FP ₂The approximate entropy nonlinear exponent be respectively 0.59 and 0.60; Whole anesthesia process nonlinear trend figure under the general anesthesia state shows that nonlinear exponent numerical value fluctuates between 0.5～0.6, numerical value keeps relative stability for a long time.This presentation of results, the approximate entropy nonlinear exponent keeps stable level in the anesthesia process, this mainly is the influence that the cerebral cortex neurocyte is subjected to anaesthetic, its active degree and contact each other have been subjected to obvious suppression, caused neurocyte electrical activity information complexity to reduce, so the numerical value of the approximate entropy of reflection EEG signals complicacy reduces.

Show from the EEG signals in each stage of anesthesia process, the EEG signals of scalp record comprises relevant consciousness cognitive function information.Each electrode of brain electricity has reflected millions of neuronic activities, particularly comes from the cones of the 5th layer in cerebral cortex.EEG signals has comprised the synchronization extent of the information about network layer, particularly localized network and the coupling of the network far away of being separated by connection situation.Compare with traditional method based on Fourier analysis, nonlinear dynamic analysis is more suitable for extracting relevant information from brain electric information.From nonlinear exponent and nonlinear trend figure, and comprehensive should being used for of judging of depth of anesthesia see, the depth of anesthesia change information that the method is more suitable for monitoring in the electrical activity of brain and is comprised.And depth of anesthesia is led the realization of the method for real-time of physiological signal more, can realize complex information that comprised in the Real Time Observation EEG signals, the activity of reflection cerebral cortex neural network.

In conjunction with the accompanying drawings the specific embodiment of the present invention has been carried out exemplary description above, obviously the present invention is not limited to this, and the various forms of changes of carrying out within the scope of the present invention all do not exceed protection scope of the present invention.

Claims (5)

1. a generation method of utilizing computer technology and nonlinear dynamic analysis method to realize being used for real-time monitored brain potential nonlinear trend figure comprises the steps:

(1) gathers original EEG signals waveform from the scalp different parts;

(2) the original EEG signals that will collect is carried out A/D and is converted to digital information, and the various interference of filtering also are stored in the storer;

(3) the operate time sequence reconstruct track of EEG signals in state space that respectively lead;

(4) in correlation dimension, complexity and three kinds of non-linear eeg dynamic analysis methods of approximate entropy, select a kind of the analysis;

It is characterized in that it also adopts following step:

(5) brain electric information that is buffered in the storer is carried out the recursion processing, reconstruct comprises the new vector set of recursion information;

(6) vector set after handling through recursion is pressed selected non-linear eeg dynamic analysis method respectively, use the method for simplifying and optimizing, carry out the calculating of correlation dimension; Select suitable and optimum parameters, carry out the calculating of complexity and approximate entropy, realize real-time analysis;

(7) utilize time series correlation dimension number, complexity and approximate entropy nonlinear exponent, reflect current nonlinear dynamic characteristic, draw out brain potential nonlinear trend figure-nonlinear trend figure in chronological order, reflect the nonlinear dynamic characteristic situation of change of whole process.

(8) store and export resulting nonlinear exponent and nonlinear trend figure.

2. the generation method of brain potential nonlinear trend figure according to claim 1 is characterized in that brain electric information is carried out recursion handles the following method of taking:

Before the track in the reconfiguration system state space, carry out recursion and handle being buffered in EEG signals V (i) in the storer, with EEG signals sequence X (i)=(x (1), x (2), x (3) ..., x (N)) handle through recursion after, convert new vector set to:

$V\left(i\right)=\left[\begin{array}{c}X(i-k)\\ X(i-k+1)\\ \·\·\·\\ X\left(i\right)\end{array}\right]$

K=m/N wherein, 0≤k≤i

Wherein m is a recursion length, is through the EEG signals sequence after the recursion processing in each data of handling, and has so just guaranteed the continuity of result; After disposing, the V (i-1) that the original EEG signals sequence of the V in the buffer memory (i) is moved in the buffer memory locates.

3. the generation method of brain potential nonlinear trend figure according to claim 1 is characterized in that following method is taked in the calculating of correlation dimension:

By the track of time series reconstructing system in state space that is studied: realize reconstruct M dimensional vector V (i) by the method that embeds time delay, L time delay, M is for embedding dimension:

V(i)＝(x(i)，x(i+L)，x(i+2×L)，...，x(i+(M-1)×L))

The wherein calculating of correlation integral:

$C\left(r\right)=\frac{1}{N_{\mathrm{ref}}}\underset{i=1}{\overset{N_{\mathrm{ref}}}{\mathrm{\Σ}}}\frac{1}{N-i}\underset{j>i}{\overset{N}{\mathrm{\Σ}}}\mathrm{\θ}(r-||V\left(j\right)-V\left(i\right)|\left|\right)$

Wherein θ is the Heavside function: if x＜0, θ (x)=0; If x 〉=0, θ (x)=1; N _RefNumber for reference point; Find out the linear segment between logC (r)/log (r), obtain the slope of this partial fitting straight line, this slope is correlation dimension D ₂:

$D_2=\underset{r\→0}{\lim }\frac{\log C\left(r\right)}{\log \left(r\right)}$

4. the generation method of brain potential nonlinear trend figure according to claim 1 is characterized in that following method is taked in the calculating of complexity and approximate entropy:

The calculating of complexity is calculated according to the algorithm of Kasper and Schuster;

The calculating of approximate entropy is calculated according to the algorithm of Pincus;

Wherein, the computational data length of complexity and approximate entropy is taked the 250-1000 data point, and selects frontal temporal part to lead and carry out real-time analysis.

5. the generation method of brain potential nonlinear trend figure according to claim 1 is characterized in that the generation of nonlinear exponent and the drafting of nonlinear trend figure take following method:

The correlation dimension that calculates as stated above, complexity and approximate entropy are according to the recursion disposal route, for R as a result _iHave:

Wherein, the span of n is between m/N～50;

Go out brain potential nonlinear trend figure by nonlinear exponent and time relation real-time rendering again.

Images